Matters Arising

Nature Vol. 288 13 November 1980 193 MATTERS ARISING

Origin of mammal-like reptiles hinge, or line of weakness. As Heaton8 has carefully avoided any taxonomic impli noted, this often happened in captor cations at all, and indeed the possibility of KEMP 1 has recently hypothesized that hinomorphs as well, where there was a a primary sister-group relationship "the synapsid [mammal-like reptile] skull solid wide suture between the skull roof between romeriids and synapsids is evolved from a limnoscelid-like skull and cheek. The only known specimen of nowhere excluded. rather than from that of a romeriid, and Romeriscus, a possible limnoscelid on The exact nature of the connection that the temporal fenestra of the mammal which Kemp placed much emphasis, is so between the skull table and cheek of like reptiles evolved directly from the poorly and incompletely preserved that it Limnoscelis is difficult to determine from remnant of the crossopterygian hinge cannot be used as the basis of a major a single, damaged skull. Even if the line". This disagrees with the widely hypothesis.9 'hinge-line' is merely the consequence of accepted theory that the Protorothyridi Hypotheses on the origin of the Synap distortion, as Reisz and Heaton now dae (Romeriidae), a group of small sida based on the development of the claim, the similar manner of breakage of reptiles with unfenestrated skull roofs, temporal fenestrae 10 are inappropriate the two sides of the skull, between the include the ancestors or near-ancestors of because these are not restricted to them table and the cheek, suggests nevertheless the Pelycosauria, the earliest synapsid or (lateral temporal fenestrae are present in that this line was weak. So also does the 2 1 mammal-like reptiles -4. Both Kemp and mesosaurs, bolosaurs, diapsids and some fact that the suture between the post 5 6 Reisz · have recognized that the pro millerosaurs) and because fenestration is orbital and squamosal is continued torothyridid skull is advanced in some only one aspect of the development of the posteriorly in a straight line by the suture cranial features compared with pely synapsid skull. Any hypothesis of the ori between the supratemporal and cosaurs and, hence, could not be ancestral gin of the synapsid reptiles should be squamosal to the hind margin of the skull. to the primitive pattern of early synapsids. linked to the structural innovations seen in My contention that the synapsid temporal Kemp's hypothesis replacing a protoro the braincase, occiput, adductor chamber, fenestra appeared within this suture line thyridid ancestor with a "limnoscelid mandibles and dentition as a whole, remains unaffected, even if the contact like" ancestor also seems invalid. because these changes gave rise to a between the respective bones was rather Kemp uses the common occurrence of a morphological complex that is unique firmer than supposed. There would still large supratemporal that contacts the among higher vertebrates. have been a thin layer of connective tissue postorbital, and a large tabular that between the bones, as in the case of any contacts the paraoccipital process of the R.R. REISZ suture, which could have expanded, opisthotic, as the basis of his hypothesis. M. J. HEATON eventually forming the temporal These, however, are primitive tetrapod aponeurosis. characters present in virtually all primitive Erindale College, amphibians and reptiles. Thus they have University of Toronto, T.S. KEMP no phylogenetic significance in the context Mississauga, Ontario, of synapsid ongms. The proper Canada L5L 1C6 The Zoological Collections, methodology for developing a theory of University Museum, phylogenetic relationships of any group of Parks Road, I. Kemp, T. S. Naturel83, 378-380 (I 980). organisms is to study the distribution 2. Carroll, R. L., J. Pal,ont. 43, 151-170 (1969). Oxford OX1 3PW, UK pattern of morphological characters 3. Clark, J. & Carroll. R. L. Bull. Mu,, comp. Zoo/. Haro. 144, 353-407 (1973). among a whole range of generally similar 4. Carroll, R. L. Forma et Functlo. 3, 165-178 (1970). organisms so that a determination of S. Reisz, R.R. Univ. Kansas Paleont. Contrib., Vertebrata (in whether a particular character is primitive the press). Growth rings 6. Reisz, R. R. in The Te"tstrial Environment and the Origin or derived may be made (only derived of Land Vertebrates (ed. Panchen, A. L.) (Academic, in dinosaur teeth characters are of any significance when London. in the press). 7. Heaton, M. J. in The Te"estrial Environment and the Origin John trying to deduce phylogenetic relation of Land Vertebrat,s (ed. Panchen, A. L.) (Academic, THERE are several difficulties with ships). Because Kemp has failed to do this, London, in the J)ress). ston's interpretation of the growth rings 8. Heaton, M. J. Bull. Okla. gtol. Sun;, 127, 1-84 (1979). 1 he has observed in dinosaur teeth • The he has been badly misled by phylo 9. Baird, D. & carroll, R. L. Science 157, 56-59 (1967). genetically insignificant primitive charac 10. Reisz, R. R. Bull. Mus. comp. Zoo/. Harv. 1'3, 27-o2 first problem is whether the rings are ters. (1972). annual. Such growth rings are found in The thesis that the synapsid temporal living and fossil ectothermic vertebrates, fenestrae evolved directly from a remnant but in the bones, not the teeth, which are of the crossopterygian hinge-line between KEMP REPLIES-Reisz and Heaton's replaced frequently in most non-mam 2 the skull roof and cheek is untenable. criticism of the use of primitive characters malian vertebrates • The teeth of cro There is no evidence that any true coty to demonstrate phylogenetic relationships codiles are replaced at intervals of 8-16 losaur, including seymouriamorphs or ignores my specific disclaimer that I was months3, making it impossible for more diadectomorphs (limnoscelids belong to proposing any particular relationship than two annual rings to occur within any 7 the latter group ) or primitive reptiles, between limnoscelids and synapsids. My given tooth. There is no evidence that ever had a cardinal line of weakness (or object was simply to suggest that the replacement ceases during adult life, nor hinge) between the skull roof and cheek in character-state of the temporal region of that the rate was substantially different in the living animals. The presence of such a the skull of the direct ancestor of synap fossil crocodilians. Data on the manner feature in the above animals is an old sids resembled that retained in the and rate of tooth replacement for dinos misconception based on the fact that post limnoscelids, for example, and not that of aurs are limited, although evidence depositional crushing tends to concentrate romeriid or protorothyridid reptiles. Of suggests multiple replacement and rapid 2 forces along the sharp angle between the course any hypothesis about the phylo turnover • skull roof and cheek leading to breakage genetic relationships of synapsids to Dinosaurs and crocodiles would be of the bone or suture. The only known particular tetrapod groups would require expected to share some similarities in skull of Limnoscelis is compressed a comprehensive study of the distribution hard-tissue deposition and to differ from dorsolaterally so that the bone has broken of derived characters of the whole mammals, for the ancestors of archosaurs in this region to give the impression of a skeleton. It was for this reason that I and mammals were separated perhaps as